In Conclusion

This concluding chapter synthesises the key methodological and conceptual contributions of Statistics in Corpus Linguistics Research , emphasising the book’s distinctive approach to making statistical reasoning accessible and meaningful for linguistic research. The book argues the primary challenge in inferential statistics lies not in mathematical complexity but in the counter-intuitive nature of probabilistic reasoning, which requires researchers to think beyond sequential human experience to consider distributions of possible events. The chapter highlights the book’s pedagogical stance of prioritising confidence intervals over traditional significance testing, demonstrating how visualising data with intervals enables immediate identification of significant differences while maintaining focus on linguistic interpretation rather than mechanical statistical procedures. This approach repositions significance testing as secondary to data visualisation and domain-specific theoretical engagement. Key methodological contributions are reviewed, including resolution of longstanding disputes in Binomial inferential statistics, retirement of inferior methods, and introduction of closed-form confidence interval formulas for complex statistical quantities using the MOVER method. The book also addresses two key issues in corpus linguistics: applying statistical methods to manual data review processes and text-clustering effects in corpus sampling, which violates random sampling assumptions and requires adjusted variance estimates. The discussion emphasises that statistical methods cannot compensate for poor experimental design and should be viewed as tools for identifying patterns rather than establishing causality. Readers are enjoined to think critically and be honest about defects in research.